Testing SOLiD vs Illumina

[tlitman]

We have tested SOLiD against Illumina for microRNA quantitation.
Here is my conclusion:

1. Both systems apparently measure the amount of miRNAs quantitatively
2. The SOLiD system outputs 30M sequences per run (high density)
3. The Illumina system outputs 6M sequences per run (standard density)
4. The SOLiD system outputs a number of "isomiRs", which in our analysis are spurious reads due to "overfitting" to the reference genome
5. The Illumina data are easier to analyse (no "color-space" issue)

Recommendation:
If ease of analysis is of importance, Illumina appears as the first choice.
If sequence output is of importance, SOLiD has the lead.

[andibody]

Are you using a GAI or II?
30M with the Solid? wow... I assume you can get this 8x per run?
How long does it take to perform a run from RNA to aligned sequence with the Solid?

[tlitman]

GAII and SOliD

Illumina: GA II.

SOLiD:
The 30M are with high-density beads.
We got 15M sequences with "normal" density beads.
Yes, this was from a single "channel" run.
The timing... hmm, it took several weeks before we got the data, but in principle, it should not take more than a week to run and analyze this "small" (2 samples) experiment.

Are you using a GAI or II?
30M with the Solid? wow... I assume you can get this 8x per run?
How long does it take to perform a run from RNA to aligned sequence with the Solid?

[Chipper]

I would guess you got a quarter for the SOLiD (it does not hav channels or lanes but can be divided into quads or octet). Or you could use barcodes of course (currently 16 per spot).

Run times are said to be 3.5 days with the 3.0 version (35 bases).

Have you decided yet if and what to buy?

[tlitman]

Correct, an octet, or rather: 2, one for each sample.

Decisions: I am inclined towards the Illumina system, mainly because it is easier to analyze the data, and because the running cost is lower.
both systems can do barcoding, although we did not barcode for this particular experiment (but for another run).

I would guess you got a quarter for the SOLiD (it does not hav channels or lanes but can be divided into quads or octet). Or you could use barcodes of course (currently 16 per spot).

Run times are said to be 3.5 days with the 3.0 version (35 bases).

Have you decided yet if and what to buy?

[tlitman]

running costs: please clarify

Today (Thu, Nov 13 2008) I received a phone call from a person at Applied Biosystems (their sales director, Eric Norbart), who accused me for spreading inaccurate information in the public domain regarding the running cost of the SOLiD system compared to Illumina

Therefore, before I retract my statement, I would like to ask both SOLiD and Illumina users to estimate their actual running cost of the systems. That is:
Q: What is the cost per gigabase output of a 35 nt run per system?

Thank you (for letting ABI know...)

- Thomas

Correct, an octet, or rather: 2, one for each sample.

Decisions: I am inclined towards the Illumina system, mainly because it is easier to analyze the data, and because the running cost is lower.
both systems can do barcoding, although we did not barcode for this particular experiment (but for another run).

[cgb]

comparison

Don't be bullied - you are the customer ! Ive heard so much fooey over the years its a bit rich being accused of spreading false information by a vendor!

1. there are no accurate costings out there - in fact there are a number of different ways of doing it. some people don't count capex at all.

2. think 'run time', how much data per unit time do you get (that BTW - is the true definition of 'throughput') last time i looked a 'run' was not a standard fixed unit of time. so how much data can you make in 1 year or over the lifetime of the machine if run continuously?

3. It is *very* hard to do like-for-like comparisons of colour-space data and illumina data. hence any accuracy claims should be viewed in context.

4. cost per run is different question (i.e cost per gigabase). Does time to result cost anything ? or is it purely the costs of the reagents ?

[tlitman]

Thanks for the reply, cgb

We still have to decide on the HTS system.
My impression is that ABI are under pressure, as they have been very aggressive in their attempt to convince us that their system is better and cheaper than Illumina's, while the latters sales reps seem much more relaxed - and thus, pleasant - in their presentations.

Regarding time vs. cost:
We are more concerned in the actual run cost than in run time, since I envisage the bottleneck will be in the downstream bioinformatics analysis rather than in throughput of sequence data

[Chipper]

I think the barcoding will make SOLiD cometetive cost-wise since my understanding is that it has a higer sample prep cost but lower cost per Gb sequence.

Perhaps Eric could enlighten us on the actual reagent costs for a SOLiD run and library prep?

[csquared]

Eric Norbart very likely did more harm than good with that phone call but that is pretty typical middle management for you.

You ABSOLUTLEY can run an Illumina instrument less expensive than the ABI when all factors are considered. How much less expensive will depend on lots of variables. However, if you consider the specs for the upcoming solid3 and its output, you can also argue that the 'per GB' cost on the solid3 will be less than the 'per GB' cost on the Illumina.

[james hadfield]

They all cost an awful lot of money; to buy and to run. And if you are going to fork out around $500k for a machine you probably want to run it as much as possible which equals vast sums of money on kits and chemistry.
Of course much of this gets ignored as soon as you start to see the wonderful things we can do with these technologies and many people opt for a second machine quite quickly.